Low power setup - what's wrong with milliamps ? (Rocket Scream & Low Power Mode)


I am currently building an autonomous sensor node, sending temperature & humidity over a 433 mhz transmitter.
The sensor node is to be battery powered.

There is no lack of documentation for such a system.

I have sourced several cheap 8mhz/3.3v arduino pro mini off ebay, $2 items.
I currently am powering the device through 4 AA batteries in a battery holder.
I wanted to use the sleep power down mode to reduce battery payload.
It does work, going into sleep power down for 10 minutes, sending correct temp, humid, voltage.

And that's where things start being funny.

I have a fully functionnal setup, with 433mhz transmitter & dht22 wired, power led desoldered on arduino board. It sends data for 4 days before running out of juice (2500mah nimh).

I have hooked a (also cheapish, $15 this time) multimeter on battery wire, in order to have a current reading.

Found several things.

Arduino mini, full system with sensor & TX, powered through vcc pin (5v, a bit high I understand)
Arduino running (setup, and everything before sleep power down command is issued) : 18 milliamps
Arduino sleeping power down, about 8 milliamps.


Powering the same setup through RAW pin (to my knowledge, that sends the 5v provided to the internal voltage regulator).
16 mA running, 2.5 mA in sleep power down.


Board must be fried or loaded with a funny thing, or I did cause some harm desoldering the cms led ?
Let's try with a brand new arduino mini pro 3.3v out of the bag (a $2 item off ebay, also).
No sensor, no nothing, power led still attached. Naked board.

Let's boot it with the same code.
Through RAW pin, 13mA running, 5mA in sleep power down.
Through VCC pin, 32 mA running, 13mA in sleep power down.


I tried different soft ways (rocket lib & direct code) without any change noticed.

That's the moment when soldering iron & multimeter are stored, and I start the browser to write a post. This post.

What I do not get is why a loaded board eats way more, and why going through regulator ends up in less power sipped.

Next steps are :

  • A different kind of arduino mini pro ($4 off ebay this time, whoooooohoooo)
  • A two cells battery holder (in case 4 AA nimh voltage is the cause, but i did try with lower voltage by inserting a cable between battery & spring holder without any success regarding consumption)
  • A bigger battery holder.
  • Solar charging setup.
  • Self contained nuclear power plant.
  • Heavy battery setup.
  • Edit : I may have one arduino mini left in a bag. I'll try this one too...

I do not wish to build an award winning setup, I just wish I could have something that eats comething under 1 milliamp in power down, in order to be able to power it through C/R14 sized alkaline batteries.

Did anyone encounter a arduino mini clone stubbornly refusing to enjoy quiescent power levels ?


So you're feeding a 3.3v pro mini?

When you feed it through RAW, it's going through the regulator. Chip is getting 3.3v.

When you feed it through Vcc, it's going straight to the chip. Chip is getting 5v. Since IC's always draw more current when running at a higher voltage, your results seem to make sense. And you're driving the regulator the wrong way, which sometimes wastes power (some regulators are worse than others in that situation), which likely serves to make the difference larger still

And you're driving the regulator the wrong way,

That makes a lot of sense, thanks a lot for your reply.

That leaves me a tad less confused about what's going on.

I'm going to lower input voltage as a first test.

I'm still uneasy with the 2.5 milliamps drain during "sleep power down", but I may have an idea or two.

First, multimeter is calibrated with 200 mA. If not, setup does not take off (in 20 mA mode i expect internal resistance to be too big, and thus to interfere with value measured).

Secondly, my "2500 mah" labelled AA nimh cells are probably 1400mah at best. (cheap ebay purchase one more time).

So with an explanation as "why the voltage reg drains less", and a bit of brain ennefort, I * might* just be in a zone where the multimeter harms the reading, and going through 4 AA cells is not exactly wasting 2500 mah.

I will probably use a switch to 'open' the power wire across the multimeter once it has reached sleep power down mode... (as explained in http://forum.arduino.cc/index.php?topic=252005.msg1784690#msg1784690 )

When I did some testing with a 5v Nano, I couldn’t get the sleep consumption to less than 3-4mA even with the power LED disabled. I later figured this was due to the regulator use 3-4mA.

In the end I built my own arduino board with just the bare essentials.

I am not using a nano.
The nano has the usb chip onboard, and this chip can also drain precious milliamps.

I use a pro mini (5v, I though it was 3.3v model, but the voltage reg is a 5v one), with power led removed.

The mini pro clone I use has a voltage reg labelled "L05", and the layout is not as per the "official" arduino mini pro layouts I have encountered.

Maybe the voltage regulator is the culprit. I use a board with this layout :

I ordered another arduino mini with the reg "transversally" mounted, that looks like the usual layout.

As soon as I receive my two cells battery holder (both for AA and C/R14 type) I'll give a try at removing the voltage reg.

I did also try to measure current drain using the correct range on the multimeter (and a bypass for startup), I'm still reading something in the 2milliamps range in sleep power down.

This is the best I could achieve, and this is by supplying 5 volts (4 cells) through the raw pin, so going through the reg.

If I manage to have a smaller consumption, in microamps I'll be very happy. If not, the 8000 mah capacity of a pair of C/R14 alkaline batteries should last about 3 month.

If I fail to reach the femtoamps drain goal with my second sort of arduino board, maybe I'll get one of the low energy mini boards (http://www.rocketscream.com/blog/product/mini-ultra-8-mhz-arduino-compatible/), or build one from scratch.

Bumping up an (very) old thread, in order to provide the modification for the 1st board pictured above for this board as I couldn’t find anything online.


You might get some ideas from this article: Temp. And Humidity Sensor With A CR2032 For Over 1 Year! - Hackster.io

And the measurements with my crappy DMM, will get another reading tomorrow when I receive a better Uni-T UT61E

20.6ma IDLE

9.6mA IDLE

Keep in mind it's the older Atmega168 at 16Mhz, I do have 2 more boards with the same layout with the Atmega328 coming on a slow boat, will report back once I get them :slight_smile:
I think that the one I've modded above is the best bang for the buck/most mod friendly version compared to the Sparkfun/Deek-Robot clones with smaller SMD passives.
It also has a proper full sized crystal (instead of a resonator) of which you can remove and use the internal oscillator if timing isn't critical or for ultra low power, or even replace it with a 8mhz one (the 5V/16Mhz are cheaper and the 3.3V/8Mhz doesn't have as many variants)

On the Sparkfun/Deek, I don't like the weird placement of A4-A5 (and A6-A7 depending on the variant) , the blue board I've modded has all of them on the outside instead of being offset, next to the IC and everything is properly silkscreened vs that el-cheapo "BTE" (Baite?) black board with the smaller IC package and where the holes are way too close to the edge of the PCB:

For 8Mhz/3.3V, the closest I've found stock in a similar layout was the RobotDyn Pro Mini : serial pinout is reversed, the analog pins don't line up and it also uses the 5 pins voltage regulator vs the SOT23-3, the less pins, the easier is the removal, although this LDO does appears to have a lower drop out vs the cheaper "L05" (78L05) SOT23-3 :slight_smile:

I currently am powering the device through 4 AA batteries in a battery holder.

Arduino mini, full system with sensor & TX, powered through vcc pin (5v, a bit high I understand)

Did you really try to power the Arduino directly from 4 batteries? Then you risk to damage the controller, by a voltage higher than 6V!

I'd run such a project directly from 2 batteries, at about 3.3V, if your transmitter also works on 3.3V, or from 3 batteries at about 4.5V.

Also check whether the voltage regulators support feeding from the output pin. Some regulators don't support that, and you cannot know which parts of what quality or with fake labels are mounted on the cheap clones.

The thread is 2 years old, so the OP must have found a solution already.
I bumped up the thread in case someone was looking for info on how to modify that paticular board.

If using alkalines yes, 4xAA would need a (ultra) low dropout regulator with a maximum drop out of 1V as Alkalines has like 1.1-1.3V through their usable range and are depleted at 1V

3xNiMh rechargeables would be perfect at 3.6V without a regulator

Or a single lithium-ion/lipo cell at 3.7V nominal